webscraping_ai_questionLABSCO SUMMARY
The question tool is the one that saves tokens, not just typing.
Pulling a whole page into the conversation to find one number costs context on every subsequent turn of that conversation. Asking the scraping service the question means only the answer comes back. Reach for fields when you need structure across many pages, selectors when you know the markup and it is stable, and html only when the other four have failed — that order is roughly the order of increasing cost to you.
What it is
WebScraping.AI's endpoint. The tools differ by what you want back, which is the useful axis — a question, structured fields, or the markup.
What you get
- Question — ask a question about the page and get the answer
- Fields — extract named fields into structure
- Text and html — the readable text, or the raw markup
- Selected and selected_multiple — the content at a CSS selector, one or many
- Account — your usage against the plan
Requirements
On the hosted endpoint, a WebScraping.AI account, signed in over OAuth; there is no key to copy. Running the open-source server locally with npx takes an API key instead, set as WEBSCRAPING_AI_API_KEY. Calls draw on the same usage-based API quota as their REST product, so the endpoint spends the credits you already buy.
Setup effort
Paste a URL, then authorize — add the endpoint to your client, then approve the OAuth consent screen
USE CASES
01
Getting one fact off a page
Ask the question; only the answer enters your context.
02
Building a table from many pages
Named fields extracted into structure.
03
Targeting known markup
One selector or many, when the page shape is stable.
04
Watching the quota
It draws on the same credits as the REST API.
7 tools
Listed in project docs. Your client may show a different list.
webscraping_ai_fieldswebscraping_ai_htmlwebscraping_ai_textwebscraping_ai_selectedwebscraping_ai_selected_multiplewebscraping_ai_accountBEFORE YOU INSTALL
What to know first
Tools you get
7 tools named in the project's documentation.
Requires
A WebScraping.AI account; hosted: an MCP client that can reach a remote endpoint; local: npx, and your API key in WEBSCRAPING_AI_API_KEY
Where it runs
The documented client runs the server on your machine.
INSTALLATION
Connect your client
Add the endpoint to your client, then approve the OAuth consent screen.
Server setup
ENDPOINT URL
https://mcp.webscraping.ai/mcpRequirements
- OAuth — add the URL, then approve WebScraping.AI's consent screen; the client obtains the bearer token itself
MCP Clients
Claude Desktop — Local
Sign up to get a free API key — 2,000 credits a month, no credit card required.
Add this to your claude_desktop_config.json:
CLAUDE DESKTOP — LOCAL · JSON
{
"mcpServers": {
"mcp-server-webscraping-ai": {
"command": "npx",
"args": [
"-y",
"webscraping-ai-mcp"
],
"env": {
"WEBSCRAPING_AI_API_KEY": "YOUR_API_KEY_HERE",
"WEBSCRAPING_AI_CONCURRENCY_LIMIT": "5",
"WEBSCRAPING_AI_ENABLE_CONTENT_SANDBOXING": "true"
}
}
}
}
